Barometre Changement Climatique

Ce rapport contient la description et l'explicabilité d'un modèle de prédiction de la solution envisagée par des sondés face aux problématique du changement climatique

General Information

Version : 0.1

Name : Barometre Changement Climatique

Purpose : Prédire la solution au changement climatique envisagée par les sondés

Date : 2023-11-22

Contributors : Thomas Bouche

Description : Un sondage de l ADEME Et OpinionWay demande à des sondés leur avis sur la solution face aux changements climatiques. Nous faisons un modèle pour prédire la solution répondue grâces aux autres questions et caractéristiques du sondé

Source Code : https://github.com/ThomasBouche/sensibilisation_explicabilite

Git Commit : None

---

Dataset Information

Path : https://www.data.gouv.fr/fr/datasets/barometre-representations-sociales-du-changement-climatique/

Origin : Data.gouv.fr

Description : Depuis 2000, l’ADEME a initié un baromètre sur les représentations sociales du changement climatique. Cette enquête permet de dresser un tableau et de mesurer les évolutions de la place de l’environnement dans les préoccupations des Français, les représentations des causes et conséquences du changement climatique, l’opinion sur les solutions et mesures de politiques publiques ainsi que l’engagement individuel. Vous trouverez ici la base de données de l’ensemble des vagues de ce baromètre depuis l’année 2000.

Depth : De 2000 à 2022

Perimeter : les vagues les plus complètes sur les 10 dernières années

Target Variable : Solutions

Target Description : Solutions face au changement climatique

---

Data Preparation

Variable Filetring : Conservation des variables qui semblent clés (une quarantaine)

Individual Filtering : Suppression des individus avec trop de valeurs à vides

Missing Values : à revoir

Feature Engineering : reconstitution d une variable qui correspond à la moyenne des actions que le sondé envisage possible de faire

Path To Script : https://github.com/ThomasBouche/sensibilisation_explicabilite

---

Model Training

Used Algorithm : catboost.

Parameters Choice : paramètres par défaut

Metrics : auc

Validation Strategy : We splitted our data into train (75%) and test (25%)

Path To Script : https://github.com/ThomasBouche/sensibilisation_explicabilite

---

Model analysis

Model used : CatBoostClassifier

Library : catboost.core

Library version : 1.2.2

Model parameters :

Parameter key	Parameter value
_object	<_catboost._CatBoost object at 0x7f4af4b633d0>
_init_params	{'max_depth': 5, 'scale_pos_weight': 1.2}
_is_fitted_	True
_random_seed	0

Parameter key	Parameter value
_learning_rate	0.036913998425006866
_tree_count	1000
_n_features_in	0
_prediction_values_change	[2.0346833081487796, 3.120371255684199, 6.57377260598391, 2.567577368551938, 2.3927540534418066, 3.1781498563106574, 2.021230804942296, 6.668565883975401, 2.646171328293313, 2.608325998126067, 2.5538167033696104, 2.2766670921713823, 2.019612015813633, 2.038943666918077, 1.7562231349278974, 2.8401239217163625,...

---

Dataset analysis

Global analysis

	Training dataset	Prediction dataset
number of features	31	31
number of observations	19,862	6,621
missing values	0	0
% missing values	0	0

Univariate analysis

Feature

• cause_changement_climatique
• cause_desordre
• causes_ges_aerosols
• causes_ges_agriculture
• causes_ges_batiments
• causes_ges_energie
• causes_ges_forets
• causes_ges_industries
• causes_ges_nucleaires
• causes_ges_transports
• causes_ges_volcans
• changement_subi
• diplome
• echiquier_politiques
• futures_consequences
• genre
• impact_effet_de_serre
• limitation_changement
• locataire_proprietaire
• mediatisation
• mode_transport
• moyenne_action_individuelle
• moyenne_mesures_souhaitables
• nb_membres_foyer
• region
• situation_pro
• sympathie_ecologie
• taille_commune
• tranche_age
• type_habitat
• veracite_scientifique

cause_changement_climatique - Categorical

	Training dataset	Prediction dataset
distinct values	3	3
missing values	0	0

cause_desordre - Categorical

	Training dataset	Prediction dataset
distinct values	3	3
missing values	0	0

causes_ges_aerosols - Categorical

	Training dataset	Prediction dataset
distinct values	5	5
missing values	0	0

causes_ges_agriculture - Categorical

	Training dataset	Prediction dataset
distinct values	5	5
missing values	0	0

causes_ges_batiments - Categorical

	Training dataset	Prediction dataset
distinct values	5	5
missing values	0	0

causes_ges_energie - Categorical

	Training dataset	Prediction dataset
distinct values	5	5
missing values	0	0

causes_ges_forets - Categorical

	Training dataset	Prediction dataset
distinct values	5	5
missing values	0	0

causes_ges_industries - Categorical

	Training dataset	Prediction dataset
distinct values	5	5
missing values	0	0

causes_ges_nucleaires - Categorical

	Training dataset	Prediction dataset
distinct values	5	5
missing values	0	0

causes_ges_transports - Categorical

	Training dataset	Prediction dataset
distinct values	5	5
missing values	0	0

causes_ges_volcans - Categorical

	Training dataset	Prediction dataset
distinct values	5	5
missing values	0	0

changement_subi - Categorical

	Training dataset	Prediction dataset
distinct values	5	5
missing values	0	0

diplome - Categorical

	Training dataset	Prediction dataset
distinct values	8	8
missing values	0	0

echiquier_politiques - Categorical

	Training dataset	Prediction dataset
distinct values	7	7
missing values	0	0

futures_consequences - Categorical

	Training dataset	Prediction dataset
distinct values	4	4
missing values	0	0

genre - Categorical

	Training dataset	Prediction dataset
distinct values	2	2
missing values	0	0

impact_effet_de_serre - Categorical

	Training dataset	Prediction dataset
distinct values	3	3
missing values	0	0

limitation_changement - Categorical

	Training dataset	Prediction dataset
distinct values	5	5
missing values	0	0

locataire_proprietaire - Categorical

	Training dataset	Prediction dataset
distinct values	4	4
missing values	0	0

mediatisation - Categorical

	Training dataset	Prediction dataset
distinct values	4	4
missing values	0	0

mode_transport - Categorical

	Training dataset	Prediction dataset
distinct values	7	7
missing values	0	0

moyenne_action_individuelle - Numeric

	Training dataset	Prediction dataset
count	19,862	6,621
mean	2.41	2.4
std	1.39	1.4
min	0	0
25%	2	2
50%	3	3
75%	3.36	3.36
max	4	4

moyenne_mesures_souhaitables - Numeric

	Training dataset	Prediction dataset
count	19,862	6,621
mean	1.85	1.82
std	1.39	1.39
min	0	0
25%	0	0
50%	2.42	2.4
75%	3	3
max	4	4

nb_membres_foyer - Categorical

	Training dataset	Prediction dataset
distinct values	6	6
missing values	0	0

region - Categorical

	Training dataset	Prediction dataset
distinct values	15	15
missing values	0	0

situation_pro - Categorical

	Training dataset	Prediction dataset
distinct values	10	10
missing values	0	0

sympathie_ecologie - Categorical

	Training dataset	Prediction dataset
distinct values	5	5
missing values	0	0

taille_commune - Categorical

	Training dataset	Prediction dataset
distinct values	5	5
missing values	0	0

tranche_age - Categorical

	Training dataset	Prediction dataset
distinct values	6	6
missing values	0	0

type_habitat - Categorical

	Training dataset	Prediction dataset
distinct values	4	4
missing values	0	0

veracite_scientifique - Categorical

	Training dataset	Prediction dataset
distinct values	3	3
missing values	0	0

Target analysis

INFO:matplotlib.category:Using categorical units to plot a list of strings that are all parsable as floats or dates. If these strings should be plotted as numbers, cast to the appropriate data type before plotting.
INFO:matplotlib.category:Using categorical units to plot a list of strings that are all parsable as floats or dates. If these strings should be plotted as numbers, cast to the appropriate data type before plotting.

solutions - Categorical

	Training dataset	Prediction dataset
distinct values	2	2
missing values	0	0

Multivariate analysis

---

Model explainability

Note : the explainability graphs were generated using the test set only.

Global feature importance plot

Feature

• cause_changement_climatique
• cause_desordre
• causes_ges_aerosols
• causes_ges_agriculture
• causes_ges_batiments
• causes_ges_energie
• causes_ges_forets
• causes_ges_industries
• causes_ges_nucleaires
• causes_ges_transports
• causes_ges_volcans
• changement_subi
• diplome
• echiquier_politiques
• futures_consequences
• genre
• impact_effet_de_serre
• limitation_changement
• locataire_proprietaire
• mediatisation
• mode_transport
• moyenne_action_individuelle
• moyenne_mesures_souhaitables
• nb_membres_foyer
• region
• situation_pro
• sympathie_ecologie
• taille_commune
• tranche_age
• type_habitat
• veracite_scientifique

Features contribution plots

cause_changement_climatique -

cause_desordre -

causes_ges_aerosols -

causes_ges_agriculture -

causes_ges_batiments -

causes_ges_energie -

causes_ges_forets -

causes_ges_industries -

causes_ges_nucleaires -

causes_ges_transports -

causes_ges_volcans -

changement_subi -

diplome -

echiquier_politiques -

futures_consequences -

genre -

impact_effet_de_serre -

limitation_changement -

locataire_proprietaire -

mediatisation -

mode_transport -

moyenne_action_individuelle -

moyenne_mesures_souhaitables -

nb_membres_foyer -

region -

situation_pro -

sympathie_ecologie -

taille_commune -

tranche_age -

type_habitat -

veracite_scientifique -

---

Model performance

Univariate analysis of target variable

INFO:matplotlib.category:Using categorical units to plot a list of strings that are all parsable as floats or dates. If these strings should be plotted as numbers, cast to the appropriate data type before plotting.
INFO:matplotlib.category:Using categorical units to plot a list of strings that are all parsable as floats or dates. If these strings should be plotted as numbers, cast to the appropriate data type before plotting.

solutions - Categorical

	True values	Prediction values
distinct values	2	2
missing values	0	0

Metrics

AUC : 0.635

---